I am not very experienced with product administration so at times I find troubleshooting problems a bit difficult.

I have an alarm setup to email me a notification if the CPU on my Unix server hits 95%. I got that email yesterday so I logged into the Reporting Server Console to check what was running by viewing the Agents. There were approx 8 deferred agents. Each of which eventually completed on their own.

At some point between getting the email alarm & logging into the server, the CPU utilization jumped to 200% & has been hovering there since. At any given time during the day I can monitor agents & see nothing all the while the CPU still appears to be pegged.

from "TOP" command:
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
3176 iadmin 18 0 1244m 1.0g 35m S 199 26.1 3257:14 java

This happened twice back in April & I was not able to find the problem. The only resolution I had was to bounce Tomcat.

I know the first logical question is "What reports are running?" I struggle with that question b/c by the time I get the emailed alarm & log in, many report requests have come & gone. I only know to view Agents when looking for running reports. Is that the correct place to look?

Has anyone else experienced this issue?

We experienced this (not frequently, but more often than occasionally) and never quite figured out why. (It liked to happen at the most inopportune times, like month end, so nothing else could run) Anyway, it seemed to be that the Java process would just take over the server. When you bounce the web/app server, it reinitializes the java process so it freed up the CPU and memory. We lived with it for a year until we moved to a recent version of WebSphere on Linux. The problem went away completely. Something with the version of Java being used and the version of the app server. It was a mystery to us as well.

I'd have to guess that you had a GRAPH request running which took some time to render. Even with that it would be hard to pinpoint as you'd actually have to know what variables were passed that yielded such high CPU.

According to your 'top' command the java process is running as the 'iadmin' user. Do you start the tomcat server with the 'iadmin' user?

It also sounds like you are running the WebFocus web client and the Reporting Server on the same Unix host. Is that correct?

Assuming it is tomcat that is the problem, it is probably garbage collection that is causing your excessive cpu usage. Do you know if you run the tomcat server with verbose garbage collection enabled?

You can tell by looking in the catalina.out file and look for lines that look like this:






= 32), weak 0, final 96, phantom 0>


If you do not see this you can enable verbose GC by specifying this in your startup.sh file:

CATALINA_OPTS="-verbose:gc -Xms512m -Xmx1024m $CATALINA_OPTS"
export CATALINA_OPTS

Hope this helps.